Development of an Intelligent Thesis Retrieval System Using AI for Educational Institutions

Abstract

This research aims to develop an intelligent thesis search system using artificial intelligence (AI) for educational institutions, to test the system's effectiveness, and to evaluate user satisfaction. The system consists of three main parts: a thesis data storage and management system using a MySQL database, PDF file uploading with embedding for data conversion into semantic search vectors, and an AI chatbot for answering questions using Natural Language Processing (NLP). The system was developed as a web application using PHP and phpMyAdmin, offering various search tools such as title, year of publication, author, and semantic search, enabling the system to understand user queries. Experimental results show that AI-powered searching is significantly more accurate than traditional systems and reduces search time. A satisfaction survey of 30 users using a 5-level satisfaction scale revealed a high level of satisfaction (average score of 4.52, standard deviation of 0.56). Most users felt the system improved research efficiency and supported learning. The system is suitable for implementation in educational institutions and significantly enhances academic knowledge management.